^{page 209 note *} The adjustments for the moments for J and U-shaped curves call for consideration. I think something might be done by assuming an exponential form. But I have not had an opportunity of going sufficiently into the point. It is clear that unadjusted moments will not always give entirely satisfactory results. The student should also remember that when the β's are large their probableerrors are large too.

^{page 214 note *} The method is given by Prof. Pearson, in Biometrika, Vol. VII, pp. 96 et seq., and the example is taken from that paper.Google Scholar